htftp.c

elinks下lynx是最重要的二个文本浏览器, 在linux下非常实用, lynx比elinks早的多, 目前好像停止开发, 这是lynx源代码

/*			File Transfer Protocol (FTP) Client
**			for a WorldWideWeb browser
**			===================================
**
**	A cache of control connections is kept.
**
** Note: Port allocation
**
**	It is essential that the port is allocated by the system, rather
**	than chosen in rotation by us (POLL_PORTS), or the following
**	problem occurs.
**
**	It seems that an attempt by the server to connect to a port which has
**	been used recently by a listen on the same socket, or by another
**	socket this or another process causes a hangup of (almost exactly)
**	one minute.  Therefore, we have to use a rotating port number.
**	The problem remains that if the application is run twice in quick
**	succession, it will hang for what remains of a minute.
**
** Authors
**	TBL	Tim Berners-lee <[e-mail]>
**	DD	Denis DeLaRoca 310 825-4580 <[e-mail]>
**	LM	Lou Montulli <[e-mail]>
**	FM	Foteos Macrides <[e-mail]>
** History:
**	 2 May 91	Written TBL, as a part of the WorldWideWeb project.
**	15 Jan 92	Bug fix: close() was used for NETCLOSE for control soc
**	10 Feb 92	Retry if cached connection times out or breaks
**	 8 Dec 92	Bug fix 921208 TBL after DD
**	17 Dec 92	Anon FTP password now just WWWuser@ suggested by DD
**			fails on princeton.edu!
**	27 Dec 93 (FM)	Fixed up so FTP now works with VMS hosts.  Path
**			must be Unix-style and cannot include the device
**			or top directory.
**	?? ??? ?? (LM)	Added code to prompt and send passwords for non
**			anonymous FTP
**	25 Mar 94 (LM)	Added code to recognize different ftp server types
**			and code to parse dates and sizes on most hosts.
**	27 Mar 93 (FM)	Added code for getting dates and sizes on VMS hosts.
**
** Notes:
**			Portions Copyright 1994 Trustees of Dartmouth College
**			Code for recognizing different FTP servers and
**			parsing "ls -l" output taken from Macintosh Fetch
**			program with permission from Jim Matthews,
**			Dartmouth Software Development Team.
*/

/*
BUGS:	@@@	Limit connection cache size!
		Error reporting to user.
		400 & 500 errors are ack'ed by user with windows.
		Use configuration file for user names

**		Note for portability this version does not use select() and
**		so does not watch the control and data channels at the
**		same time.
*/

#include <HTUtils.h>

#include <HTAlert.h>

#include <HTFTP.h>	/* Implemented here */
#include <HTTCP.h>
#include <HTTP.h>
#include <HTFont.h>

#define REPEAT_PORT	/* Give the port number for each file */
#define REPEAT_LISTEN	/* Close each listen socket and open a new one */

/* define POLL_PORTS		 If allocation does not work, poll ourselves.*/
#define LISTEN_BACKLOG 2	/* Number of pending connect requests (TCP)*/

#define FIRST_TCP_PORT	1024	/* Region to try for a listening port */
#define LAST_TCP_PORT	5999

#define LINE_LENGTH 256

#include <HTParse.h>
#include <HTAnchor.h>
#include <HTFile.h>	/* For HTFileFormat() */
#include <HTBTree.h>
#include <HTChunk.h>
#ifndef IPPORT_FTP
#define IPPORT_FTP	21
#endif /* !IPORT_FTP */

#include <LYUtils.h>
#include <LYGlobalDefs.h>
#include <LYStrings.h>
#include <LYLeaks.h>

typedef struct _connection {
    struct _connection *	next;	/* Link on list		*/
    unsigned long		addr;	/* IP address		*/
    int				socket; /* Socket number for communication */
    BOOL			binary; /* Binary mode? */
} connection;

/*		Hypertext object building machinery
*/
#include <HTML.h>

#define PUTC(c)      (*targetClass.put_character) (target, c)
#define PUTS(s)      (*targetClass.put_string)    (target, s)
#define START(e)     (*targetClass.start_element) (target, e, 0, 0, -1, 0)
#define END(e)       (*targetClass.end_element)   (target, e, 0)
#define FREE_TARGET  (*targetClass._free)         (target)
#define ABORT_TARGET (*targetClass._free)         (target)

struct _HTStructured {
	CONST HTStructuredClass *	isa;
	/* ... */
};

/*	Global Variables
**	---------------------
*/
PUBLIC int HTfileSortMethod = FILE_BY_NAME;

#ifndef DISABLE_FTP /*This disables everything to end-of-file */
PRIVATE char ThisYear[8];
PRIVATE char LastYear[8];
PRIVATE int TheDate;
PRIVATE BOOLEAN HaveYears = FALSE;

/*	Module-Wide Variables
**	---------------------
*/
PRIVATE connection * connections = NULL;/* Linked list of connections */
PRIVATE char response_text[LINE_LENGTH+1];/* Last response from ftp host */
PRIVATE connection * control = NULL;	/* Current connection */
PRIVATE int data_soc = -1;		/* Socket for data transfer =invalid */
PRIVATE char *user_entered_password = NULL;
PRIVATE char *last_username_and_host = NULL;

/*
 * ProFTPD 1.2.5rc1 is known to have a broken implementation of RETR.  If asked
 * to retrieve a directory, it gets confused and fails subsequent commands such
 * as CWD and LIST.  Since this is an unusual bug, we should remove this ifdef
 * at some point - TD 2004/1/1.
 */
#define BROKEN_PROFTPD 1
PRIVATE int ProFTPD_bugs = FALSE;

typedef enum {
	GENERIC_SERVER
	, MACHTEN_SERVER
	, UNIX_SERVER
	, VMS_SERVER
	, CMS_SERVER
	, DCTS_SERVER
	, TCPC_SERVER
	, PETER_LEWIS_SERVER
	, NCSA_SERVER
	, WINDOWS_NT_SERVER
	, WINDOWS_2K_SERVER
	, MS_WINDOWS_SERVER
	, MSDOS_SERVER
	, APPLESHARE_SERVER
	, NETPRESENZ_SERVER
	, DLS_SERVER
} eServerType;

PRIVATE eServerType server_type = GENERIC_SERVER; /* the type of ftp host */
PRIVATE int	unsure_type = FALSE;		/* sure about the type? */
PRIVATE BOOLEAN use_list = FALSE;		/* use the LIST command? */

PRIVATE int	interrupted_in_next_data_char = FALSE;

#ifdef POLL_PORTS
PRIVATE PortNumber	port_number = FIRST_TCP_PORT;
#endif /* POLL_PORTS */

PRIVATE int	master_socket = -1;	/* Listening socket = invalid	*/
PRIVATE char	port_command[255];	/* Command for setting the port */
PRIVATE fd_set	open_sockets;		/* Mask of active channels */
PRIVATE int	num_sockets;		/* Number of sockets to scan */
PRIVATE PortNumber	passive_port;	/* Port server specified for data */


#define NEXT_CHAR HTGetCharacter()	/* Use function in HTFormat.c */

#define DATA_BUFFER_SIZE 2048
PRIVATE char data_buffer[DATA_BUFFER_SIZE];		/* Input data buffer */
PRIVATE char * data_read_pointer;
PRIVATE char * data_write_pointer;
#define NEXT_DATA_CHAR next_data_char()
PRIVATE int close_connection PARAMS((
	connection *	con));


#ifdef LY_FIND_LEAKS
/*
**  This function frees module globals. - FM
*/
PRIVATE void free_FTPGlobals NOARGS
{
    FREE(user_entered_password);
    FREE(last_username_and_host);
    if (control) {
	if (control->socket != -1)
	    close_connection(control);
	FREE(control);
    }
}
#endif /* LY_FIND_LEAKS */

/* PUBLIC						HTVMS_name()
**		CONVERTS WWW name into a VMS name
** ON ENTRY:
**	nn		Node Name (optional)
**	fn		WWW file name
**
** ON EXIT:
**	returns		vms file specification
**
** Bug: Returns pointer to static -- non-reentrant
*/
PUBLIC char * HTVMS_name ARGS2(
	CONST char *,	nn,
	CONST char *,	fn)
{

/*	We try converting the filename into Files-11 syntax.  That is, we assume
**	first that the file is, like us, on a VMS node.  We try remote
**	(or local) DECnet access.  Files-11, VMS, VAX and DECnet
**	are trademarks of Digital Equipment Corporation.
**	The node is assumed to be local if the hostname WITHOUT DOMAIN
**	matches the local one. @@@
*/
    static char *vmsname;
    char * filename = (char*)malloc(strlen(fn)+1);
    char * nodename = (char*)malloc(strlen(nn)+2+1);	/* Copies to hack */
    char *second;		/* 2nd slash */
    char *last;			/* last slash */

    CONST char * hostname = HTHostName();

    if (!filename || !nodename)
	outofmem(__FILE__, "HTVMSname");
    strcpy(filename, fn);
    strcpy(nodename, "");	/* On same node?  Yes if node names match */
    if (strncmp(nn, "localhost", 9)) {
	CONST char *p;
	CONST char *q;
	for (p = hostname, q = nn;
	     *p && *p != '.' && *q && *q != '.'; p++, q++){
	    if (TOUPPER(*p) != TOUPPER(*q)) {
		char *r;
		strcpy(nodename, nn);
		r = strchr(nodename, '.');	/* Mismatch */
		if (r)
		    *r = '\0';			/* Chop domain */
		strcat(nodename, "::");		/* Try decnet anyway */
		break;
	    }
	}
    }

    second = strchr(filename+1, '/');		/* 2nd slash */
    last = strrchr(filename, '/');	/* last slash */

    if (!second) {				/* Only one slash */
	HTSprintf0(&vmsname, "%s%s", nodename, filename + 1);
    } else if (second == last) {		/* Exactly two slashes */
	*second = '\0';		/* Split filename from disk */
	HTSprintf0(&vmsname, "%s%s:%s", nodename, filename+1, second+1);
	*second = '/';	/* restore */
    } else {				/* More than two slashes */
	char * p;
	*second = '\0';		/* Split disk from directories */
	*last = '\0';		/* Split dir from filename */
	HTSprintf0(&vmsname, "%s%s:[%s]%s",
		nodename, filename+1, second+1, last+1);
	*second = *last = '/';	/* restore filename */
	for (p = strchr(vmsname, '['); *p!=']'; p++)
	    if (*p == '/')
		*p = '.';		/* Convert dir sep.  to dots */
    }
    FREE(nodename);
    FREE(filename);
    return vmsname;
}

/*	Procedure: Read a character from the data connection
**	----------------------------------------------------
*/
PRIVATE int next_data_char NOARGS
{
    int status;
    if (data_read_pointer >= data_write_pointer) {
	status = NETREAD(data_soc, data_buffer, DATA_BUFFER_SIZE);
      if (status == HT_INTERRUPTED)
	interrupted_in_next_data_char = 1;
      if (status <= 0)
	return -1;
      data_write_pointer = data_buffer + status;
      data_read_pointer = data_buffer;
    }
#ifdef NOT_ASCII
    {
	char c = *data_read_pointer++;
	return FROMASCII(c);
    }
#else
    return UCH(*data_read_pointer++);
#endif /* NOT_ASCII */
}


/*	Close an individual connection
**
*/
PRIVATE int close_connection ARGS1(
	connection *,	con)
{
    connection * scan;
    int status;
    CTRACE((tfp, "HTFTP: Closing control socket %d\n", con->socket));
    status = NETCLOSE(con->socket);
    if (TRACE && status != 0) {
#ifdef UNIX
	CTRACE((tfp, "HTFTP:close_connection: %s", LYStrerror(errno)));
#else
	if (con->socket != INVSOC)
	    HTInetStatus("HTFTP:close_connection");
#endif
    }
    con->socket = -1;
    if (connections == con) {
	connections = con->next;
	return status;
    }
    for (scan = connections; scan; scan = scan->next) {
	if (scan->next == con) {
	    scan->next = con->next;	/* Unlink */
	    if (control == con)
		control = (connection*)0;
	    return status;
	} /*if */
    } /* for */
    return -1;		/* very strange -- was not on list. */
}

PRIVATE char *help_message_buffer = NULL;  /* global :( */

PRIVATE void init_help_message_cache NOARGS
{
    FREE(help_message_buffer);
}

PRIVATE void help_message_cache_add ARGS1(
	char *,		string)
{
    if (help_message_buffer)
	StrAllocCat(help_message_buffer, string);
    else
	StrAllocCopy(help_message_buffer, string);

    CTRACE((tfp,"Adding message to help cache: %s\n",string));
}

PRIVATE char *help_message_cache_non_empty NOARGS
{
  return(help_message_buffer);
}

PRIVATE char *help_message_cache_contents NOARGS
{
   return(help_message_buffer);
}

/*	Send One Command
**	----------------
**
**	This function checks whether we have a control connection, and sends
**	one command if given.
**
** On entry,
**	control	points to the connection which is established.
**	cmd	points to a command, or is zero to just get the response.
**
**	The command should already be terminated with the CRLF pair.
**
** On exit,
**	returns:  1 for success,
**		  or negative for communication failure (in which case
**		  the control connection will be closed).
*/
PRIVATE int write_cmd ARGS1(
	char *,		cmd)
{
    int status;

    if (!control) {
	CTRACE((tfp, "HTFTP: No control connection set up!!\n"));
	return -99;
    }

    if (cmd) {
	CTRACE((tfp, "  Tx: %s", cmd));
#ifdef NOT_ASCII
	{
	    char * p;
	    for (p = cmd; *p; p++) {
		*p = TOASCII(*p);
	    }
	}
#endif /* NOT_ASCII */
	status = NETWRITE(control->socket, cmd, (int)strlen(cmd));
	if (status < 0) {
	    CTRACE((tfp, "HTFTP: Error %d sending command: closing socket %d\n",
			status, control->socket));
	    close_connection(control);
	    return status;
	}
    }
    return 1;
}

/*	Execute Command and get Response
**	--------------------------------
**
**	See the state machine illustrated in RFC959, p57. This implements
**	one command/reply sequence.  It also interprets lines which are to
**	be continued, which are marked with a "-" immediately after the
**	status code.
**
**	Continuation then goes on until a line with a matching reply code
**	an a space after it.
**
** On entry,
**	control	points to the connection which is established.
**	cmd	points to a command, or is zero to just get the response.
**
**	The command must already be terminated with the CRLF pair.
**
** On exit,
**	returns:  The first digit of the reply type,
**		  or negative for communication failure.
*/
PRIVATE int response ARGS1(
	char *,		cmd)
{
    int result;				/* Three-digit decimal code */
    int continuation_response = -1;
    int status;

    if ((status = write_cmd(cmd)) < 0)
	return status;

    do {
	char *p = response_text;
	for (;;) {
	    int ich = NEXT_CHAR;
	    if (((*p++ = (char) ich) == LF)
			|| (p == &response_text[LINE_LENGTH])) {

		char continuation;

		if (interrupted_in_htgetcharacter) {
		    CTRACE((tfp, "HTFTP: Interrupted in HTGetCharacter, apparently.\n"));
		    NETCLOSE (control->socket);
		    control->socket = -1;
		    return HT_INTERRUPTED;
		}

		*p = '\0';			/* Terminate the string */
		CTRACE((tfp, "    Rx: %s", response_text));

		/* Check for login or help messages */
		if (!strncmp(response_text,"230-",4) ||
		    !strncmp(response_text,"250-",4) ||
		    !strncmp(response_text,"220-",4))
		    help_message_cache_add(response_text+4);

		sscanf(response_text, "%d%c", &result, &continuation);
		if  (continuation_response == -1) {
			if (continuation == '-')  /* start continuation */
			    continuation_response = result;
		} else {	/* continuing */
			if (continuation_response == result &&
			    continuation == ' ')
			    continuation_response = -1; /* ended */
		}
#ifdef BROKEN_PROFTPD
		if (result == 220 && LYstrstr(response_text, "ProFTPD 1.2.5")) {
		    ProFTPD_bugs = TRUE;
		    CTRACE((tfp, "This server is broken (RETR)\n"));
		}
#endif
		break;
	    } /* if end of line */

	    if (interrupted_in_htgetcharacter) {
		CTRACE((tfp, "HTFTP: Interrupted in HTGetCharacter, apparently.\n"));
		NETCLOSE (control->socket);